Voice Active Tools for Linux - OpenSesame
PROJECT
Fiscal Host: Copper Linux User Group
To design tools with privacy and accessibilty by design using vosk-kaldi with the user input coming from microphone, converts to text then executes a command
About
This is just the beginning!
Voice activated open terminal or launch terminal using vosk, kaldi and pretrained model. For different languages, get and set path for other models.
[x] Open Terminal test
[x] Open Firefox to "website" domain (no dot)
[x] Purge systemd with confirmation
[x] Tkinter front end for open term test.
[x] Open Firefox to "website" domain (no dot)
[x] Purge systemd with confirmation
[x] Tkinter front end for open term test.
- Capture voice input using a microphone.
- Convert speech to text using a speech recognition engine.
- Process the text with NLP to determine intent.
- Execute terminal commands via subprocess.
- Ensure compatibility with init-based Linux systems
How It Works:
- Uses Vosk for offline speech recognition.
- Listens for trigger words like "open terminal" or "launch terminal".
- Opens a terminal emulator based on available shells.
- Uses init-compatible commands (avoiding systemd).
- Supports multiple terminal emulators.
- Multiple language support - just add model and edit path...
Requirements:
Install dependencies:
MX specific
sudo apt install python3-pip portaudio19-dev pip install sounddevice numpy
In general
sudo apt install vosk-api sounddevice python3-pip pip install sounddevice vosk
Get model
Small - fast, less accurate ~50MB
mkdir -p ~/vosk_model cd ~/vosk_model wget https://alphacephei.com/vosk/models/vosk-model-small-en-us-0.15.zip unzip vosk-model-small-en-us-0.15.zip
Best accuracy model: 3.8 gb
Run script
Run the script, speak into the microphone, and say "open terminal" to launch the terminal.
Our team
Marcia Wilbur
Admin
Contribute
Become a financial contributor.